Predicting EA Sports FIFA Team of the Season in Europe’s Five Leagues
Introduction
The video game FIFA, which is developed by Electronic Arts (EA) Sports, has become the most popular sports video game in the world in recent years, largely due to its game mode Ultimate Team. The objective of Ultimate Team is to build the best team possible through both buying and selling players, as well as buying packs of cards similarly to how people buy soccer trading cards in real life. Each player receives ratings in various categories based on their real life abilities, and each of these ratings factor into their overall rating. At the end of each season, EA Sports creates a Team of the Season (TOTS), where they select the best player at each position in each league from that season based on how they performed in real life. The players who receive TOTS cards also receive a boost to their overall rating to reflect their abilities in real life. Although most of their choices for TOTS are understandable, there are some choices that confuse and sometimes anger fans. Along with this, EA has never explained how they make their choices. Through the use of machine learning methods and predictive modeling, we aim to determine which variables are most important when choosing a player for TOTS, as well as predict the Team of the Season for Europe’s top five leagues based on this season’s statistics.
Methods and Materials
Materials: We retrieved complete player datasets for FIFA 17, FIFA 18, and FIFA 19 from here. We retrieved real life statistics from the 2016-2017, 2017-2018, and 2018-2019 seasons from fbref.com. We did not use data from the 2019-2020 season because COVID-19 caused each season to prematurely end in March of 2020.
Methods:
English Premier League
## Truth
## Prediction Normal TOTS
## Normal 116 7
## TOTS 9 10## Truth
## Prediction Normal TOTS
## Normal 116 7
## TOTS 9 10## Player revision position Int TklW OG PKcon Nation
## 1 Eric Dier 17 Normal CDM 37 34 0 0 ENG
## 2 Adam Lallana 17 TOTS CM 20 35 0 0 ENG
## 3 Sadio Mane 17 TOTS RW 11 18 0 1 SEN
## 4 Victor Moses 17 Normal RB 41 42 0 0 NGA
## 5 Paul Pogba 17 Normal CM 37 40 0 1 FRA
## 6 Victor Wanyama 17 Normal CDM 39 64 0 0 KEN
## 7 Philippe Coutinho 17 Normal LW 18 25 0 0 BRA
## 8 Sergio Aguero 18 TOTS ST 8 5 0 0 ARG
## 9 Eric Dier 18 Normal CB 30 35 0 0 ENG
## 10 Abdoulaye Doucoure 18 TOTS CDM 41 41 0 1 FRA
## 11 Andrew Robertson 18 TOTS LB 24 21 0 0 SCO
## 12 Antonio Valencia 18 Normal RB 43 37 0 0 ECU
## 13 Christian Eriksen 19 TOTS CAM 11 27 0 0 DEN
## 14 Harry Kane 19 Normal ST 4 7 0 0 ENG
## 15 James Maddison 19 TOTS CAM 12 34 0 0 ENG
## 16 Callum Wilson 19 Normal ST 1 9 0 0 ENG
## Squad Age Born MP Min minutes_played_divided_by90 Gls Ast
## 1 Tottenham 22 1994 36 3043 33.8 2 1
## 2 Liverpool 28 1988 31 2348 26.1 8 6
## 3 Liverpool 24 1992 27 2235 24.8 13 5
## 4 Chelsea 25 1990 34 2483 27.6 3 2
## 5 Manchester Utd 23 1993 30 2608 29.0 5 4
## 6 Tottenham 25 1991 36 3012 33.5 4 1
## 7 Liverpool 24 1992 31 2227 24.7 13 8
## 8 Manchester City 29 1988 25 1963 21.8 21 6
## 9 Tottenham 23 1994 34 2824 31.4 0 2
## 10 Watford 24 1993 37 3324 36.9 7 3
## 11 Liverpool 23 1994 22 1940 21.6 1 5
## 12 Manchester Utd 31 1985 31 2740 30.4 3 1
## 13 Tottenham 26 1992 35 2774 30.8 8 12
## 14 Tottenham 25 1993 28 2424 26.9 17 4
## 15 Leicester City 21 1996 36 2831 31.5 7 7
## 16 Bournemouth 26 1992 30 2528 28.1 14 9
## Non_PK_G PK PKatt CrdY CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90
## 1 2 0 0 6 0 0.06 0.03 0.09 0.06
## 2 8 0 0 3 0 0.31 0.23 0.54 0.31
## 3 13 0 0 4 0 0.52 0.20 0.72 0.52
## 4 3 0 0 4 0 0.11 0.07 0.18 0.11
## 5 5 0 0 7 0 0.17 0.14 0.31 0.17
## 6 4 0 0 10 0 0.12 0.03 0.15 0.12
## 7 13 0 0 2 0 0.53 0.32 0.85 0.53
## 8 17 4 4 2 0 0.96 0.28 1.24 0.78
## 9 0 0 0 4 0 0.00 0.06 0.06 0.00
## 10 7 0 0 10 0 0.19 0.08 0.27 0.19
## 11 1 0 0 2 0 0.05 0.23 0.28 0.05
## 12 3 0 0 7 0 0.10 0.03 0.13 0.10
## 13 8 0 0 3 0 0.26 0.39 0.65 0.26
## 14 13 4 4 5 0 0.63 0.15 0.78 0.48
## 15 6 1 2 4 1 0.22 0.22 0.45 0.19
## 16 13 1 2 3 0 0.50 0.32 0.82 0.46
## G_plus_A_minus_PK_per90 Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS
## 1 0.09 2 86 26 60 86 31639 0.0002857143 0.99971429
## 2 0.54 4 78 42 36 76 53016 0.9690373056 0.03096269
## 3 0.72 4 78 42 36 76 53016 0.7454276848 0.25457232
## 4 0.18 1 85 33 52 93 41508 0.0019047619 0.99809524
## 5 0.31 6 54 29 25 69 75290 0.2535569210 0.74644308
## 6 0.15 2 86 26 60 86 31639 0.0187142857 0.98128571
## 7 0.85 4 78 42 36 76 53016 0.2255183441 0.77448166
## 8 1.05 1 106 27 79 100 54070 0.7623181004 0.23768190
## 9 0.06 3 74 36 38 77 67953 0.0850712432 0.91492876
## 10 0.27 14 44 64 -20 41 20231 0.9763224276 0.02367757
## 11 0.28 4 84 38 46 75 53049 0.9166878037 0.08331220
## 12 0.13 2 68 28 40 81 74976 0.0745622120 0.92543779
## 13 0.65 4 67 39 28 71 54216 0.8036522291 0.19634777
## 14 0.63 4 67 39 28 71 54216 0.3609930723 0.63900693
## 15 0.41 9 51 48 3 52 31851 0.9491904589 0.05080954
## 16 0.78 14 56 70 -14 45 10532 0.2149315094 0.78506849
## .pred_class
## 1 TOTS
## 2 Normal
## 3 Normal
## 4 TOTS
## 5 TOTS
## 6 TOTS
## 7 TOTS
## 8 Normal
## 9 TOTS
## 10 Normal
## 11 Normal
## 12 TOTS
## 13 Normal
## 14 TOTS
## 15 Normal
## 16 TOTS## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Harry Kane Normal ST 13 10 0 0 ENG Tottenham 27
## 2 Mohamed Salah Normal RW 6 13 0 0 EGY Liverpool 28
## 3 Timo Werner Normal ST 6 15 0 0 GER Chelsea 25
## 4 Ollie Watkins Normal ST 9 20 0 0 ENG Aston Villa 25
## 5 Jamie Vardy Normal ST 7 8 0 0 ENG Leicester City 34
## Born MP Starts Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt
## 1 1993 30 30 2632 29.2 21 13 17 4 4
## 2 1992 32 29 2633 29.3 20 3 14 6 6
## 3 1996 31 25 2243 24.9 6 6 6 0 0
## 4 1995 32 32 2880 32.0 12 4 11 1 2
## 5 1987 29 26 2401 26.7 13 8 7 6 7
## CrdY CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90
## 1 1 0 0.72 0.44 1.16 0.58
## 2 0 0 0.68 0.10 0.79 0.48
## 3 1 0 0.24 0.24 0.48 0.24
## 4 2 0 0.37 0.12 0.50 0.34
## 5 1 0 0.49 0.30 0.79 0.26
## G_plus_A_minus_PK_per90 Matches Rk GF GA GD Pts Attendance .pred_Normal
## 1 1.03 Matches 7 56 38 18 53 125 0.1230343
## 2 0.58 Matches 6 55 39 16 54 353 0.2559387
## 3 0.48 Matches 4 51 31 20 58 125 0.4411583
## 4 0.47 Matches 11 46 37 9 45 NA 0.4659519
## 5 0.56 Matches 3 60 38 22 62 NA 0.5636206
## .pred_TOTS .pred_class
## 1 0.8769657 TOTS
## 2 0.7440613 TOTS
## 3 0.5588417 TOTS
## 4 0.5340481 TOTS
## 5 0.4363794 Normal## Player revision position Int TklW OG PKcon Nation Squad
## 1 Rodri Normal CDM 31 54 0 0 ESP Manchester City
## 2 Bruno Fernandes Normal CAM 18 36 0 1 POR Manchester Utd
## 3 Son Heung min Normal LM 20 12 0 0 KOR Tottenham
## 4 Marcus Rashford Normal LM 10 7 0 0 ENG Manchester Utd
## 5 Mason Mount Normal CAM 34 44 0 0 ENG Chelsea
## Age Born MP Starts Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt
## 1 24 1996 29 27 2353 26.1 2 1 1 1 1
## 2 26 1994 33 32 2821 31.3 16 11 8 8 9
## 3 28 1992 32 31 2665 29.6 15 9 14 1 1
## 4 23 1997 33 31 2686 29.8 10 8 10 0 0
## 5 22 1999 32 28 2545 28.3 6 4 5 1 1
## CrdY CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90
## 1 4 0 0.08 0.04 0.11 0.04
## 2 5 0 0.51 0.35 0.86 0.26
## 3 0 0 0.51 0.30 0.81 0.47
## 4 4 0 0.34 0.27 0.60 0.34
## 5 2 0 0.21 0.14 0.35 0.18
## G_plus_A_minus_PK_per90 Matches Rk GF GA GD Pts Attendance .pred_Normal
## 1 0.08 Matches 1 69 24 45 77 NA 0.1087317
## 2 0.61 Matches 2 64 35 29 67 NA 0.1416100
## 3 0.78 Matches 7 56 38 18 53 125 0.2159629
## 4 0.60 Matches 2 64 35 29 67 NA 0.2218100
## 5 0.32 Matches 4 51 31 20 58 125 0.3730710
## .pred_TOTS .pred_class
## 1 0.8912683 TOTS
## 2 0.8583900 TOTS
## 3 0.7840371 TOTS
## 4 0.7781900 TOTS
## 5 0.6269290 TOTS## Player revision position Int TklW OG PKcon Nation Squad
## 1 Harry Maguire Normal CB 60 18 0 0 ENG Manchester Utd
## 2 Aaron Wan Bissaka Normal RB 63 48 0 0 ENG Manchester Utd
## 3 Ruben Dias Normal CB 21 19 1 1 POR Manchester City
## 4 Luke Shaw Normal LB 23 39 1 0 ENG Manchester Utd
## 5 Matt Targett Normal LB 37 46 0 1 ENG Aston Villa
## Age Born MP Starts Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt
## 1 28 1993 33 33 2970 33.0 2 1 2 0 0
## 2 23 1997 31 31 2790 31.0 2 2 2 0 0
## 3 23 1997 29 29 2573 28.6 1 0 1 0 0
## 4 25 1995 29 27 2384 26.5 1 5 1 0 0
## 5 25 1995 32 32 2864 31.8 0 1 0 0 0
## CrdY CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90
## 1 10 0 0.06 0.03 0.09 0.06
## 2 3 0 0.06 0.06 0.13 0.06
## 3 3 0 0.03 0.00 0.03 0.03
## 4 7 0 0.04 0.19 0.23 0.04
## 5 7 0 0.00 0.03 0.03 0.00
## G_plus_A_minus_PK_per90 Matches Rk GF GA GD Pts Attendance .pred_Normal
## 1 0.09 Matches 2 64 35 29 67 NA 0.1639413
## 2 0.13 Matches 2 64 35 29 67 NA 0.1932982
## 3 0.03 Matches 1 69 24 45 77 NA 0.2619920
## 4 0.23 Matches 2 64 35 29 67 NA 0.3116461
## 5 0.03 Matches 11 46 37 9 45 NA 0.3983746
## .pred_TOTS .pred_class
## 1 0.8360587 TOTS
## 2 0.8067018 TOTS
## 3 0.7380080 TOTS
## 4 0.6883539 TOTS
## 5 0.6016254 TOTS
La Liga (Spain)
## Truth
## Prediction Normal TOTS
## Normal 114 6
## TOTS 4 9## Truth
## Prediction Normal TOTS
## Normal 114 6
## TOTS 4 9## Player revision position Int TklW OG PKcon Nation
## 1 Sergi Roberto 17 Normal RB 49 44 0 0 ESP
## 2 Kevin Prince Boateng 17 TOTS ST 19 16 0 2 GHA
## 3 Dani Carvajal 17 TOTS RB 45 41 0 0 ESP
## 4 Karim Benzema 18 Normal ST 6 6 0 0 FRA
## 5 Koke 18 Normal CM 23 41 0 0 ESP
## 6 Marcelo 18 Normal LB 26 32 0 0 BRA
## 7 Roberto 18 TOTS RB 0 0 0 0 ESP
## 8 Ever Banega 19 TOTS CDM 31 44 0 1 ARG
## 9 Djene 19 TOTS CB 59 36 0 3 TOG
## 10 Mario Hermoso 19 TOTS CB 25 25 0 2 ESP
## Squad Age Born MP Min minutes_played_divided_by90 Gls Ast
## 1 Barcelona 24 1992 32 2385 26.5 0 6
## 2 Las Palmas 29 1987 28 1978 22.0 10 4
## 3 Real Madrid 24 1992 23 2018 22.4 0 4
## 4 Real Madrid 29 1987 32 2149 23.9 5 9
## 5 Atl\xe9tico Madrid 25 1992 35 2753 30.6 4 3
## 6 Real Madrid 29 1988 28 2262 25.1 2 6
## 7 M\xe1laga 31 1986 34 3060 34.0 0 0
## 8 Sevilla 30 1988 32 2667 29.6 3 5
## 9 Getafe 26 1991 34 2976 33.1 0 0
## 10 Espanyol 23 1995 32 2806 31.2 3 0
## Non_PK_G PK PKatt CrdY CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90
## 1 0 0 0 5 0 0.00 0.23 0.23 0.00
## 2 10 0 0 11 3 0.46 0.18 0.64 0.46
## 3 0 0 0 11 0 0.00 0.18 0.18 0.00
## 4 3 2 2 0 0 0.21 0.38 0.59 0.13
## 5 4 0 0 3 0 0.13 0.10 0.23 0.13
## 6 2 0 0 3 1 0.08 0.24 0.32 0.08
## 7 0 0 0 0 0 0.00 0.00 0.00 0.00
## 8 1 2 2 17 2 0.10 0.17 0.27 0.03
## 9 0 0 0 13 2 0.00 0.00 0.00 0.00
## 10 3 0 0 7 0 0.10 0.00 0.10 0.10
## G_plus_A_minus_PK_per90 Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS
## 1 0.23 2 116 37 79 90 78034 0.4945020 0.5054980
## 2 0.64 14 53 74 -21 39 20249 0.8094432 0.1905568
## 3 0.18 1 106 41 65 93 69426 0.7078319 0.2921681
## 4 0.50 3 94 44 50 76 66161 0.4543076 0.5456924
## 5 0.23 2 58 22 36 79 55483 0.3289353 0.6710647
## 6 0.32 3 94 44 50 76 66161 0.4382130 0.5617870
## 7 0.00 20 24 61 -37 20 20420 0.8628984 0.1371016
## 8 0.20 6 62 47 15 59 35993 0.7037185 0.2962815
## 9 0.00 5 48 35 13 59 11000 0.7566533 0.2433467
## 10 0.10 7 48 50 -2 53 19388 0.9327282 0.0672718
## .pred_class
## 1 TOTS
## 2 Normal
## 3 Normal
## 4 TOTS
## 5 TOTS
## 6 TOTS
## 7 Normal
## 8 Normal
## 9 Normal
## 10 Normal## Player revision position Int TklW OG PKcon Nation
## 1 Lionel Messi Normal RW 6 12 0 0 ARG
## 2 Karim Benzema Normal ST 11 5 0 0 FRA
## 3 Luis Suarez Normal ST 5 4 0 0 URU
## 4 Antoine Griezmann Normal ST 11 23 0 0 FRA
## 5 Mikel Oyarzabal Normal LW 16 13 0 0 ESP
## Squad Age Born MP Min minutes_played_divided_by90 Gls Ast
## 1 Barcelona 33 1987 30 2573 28.6 25 9
## 2 Real Madrid 33 1987 29 2458 27.3 21 7
## 3 Atl\xe9tico Madrid 34 1987 27 2078 23.1 19 2
## 4 Barcelona 30 1991 30 2095 23.3 11 6
## 5 Real Sociedad 24 1997 28 2015 22.4 10 7
## Non_PK_G PK PKatt CrdY CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90
## 1 22 3 4 4 0 0.87 0.31 1.19 0.77
## 2 20 1 1 2 0 0.77 0.26 1.03 0.73
## 3 16 3 3 5 0 0.82 0.09 0.91 0.69
## 4 10 1 2 4 0 0.47 0.26 0.73 0.43
## 5 4 6 7 1 0 0.45 0.31 0.76 0.18
## G_plus_A_minus_PK_per90 Matches Rk GF GA GD Pts Attendance .pred_Normal
## 1 1.08 Matches 3 76 29 47 71 NA 0.1662030
## 2 0.99 Matches 2 56 24 32 71 NA 0.3327133
## 3 0.78 Matches 1 60 22 38 73 NA 0.4027990
## 4 0.69 Matches 3 76 29 47 71 NA 0.4755379
## 5 0.49 Matches 5 51 34 17 53 NA 0.5653534
## .pred_TOTS .pred_class
## 1 0.8337970 TOTS
## 2 0.6672867 TOTS
## 3 0.5972010 TOTS
## 4 0.5244621 TOTS
## 5 0.4346466 Normal## Player revision position Int TklW OG PKcon Nation Squad
## 1 Marcos Llorente Normal CM 27 41 0 0 ESP Atl\xe9tico Madrid
## 2 Angel Correa Normal RM 19 26 0 0 ARG Atl\xe9tico Madrid
## 3 Koke Normal CM 32 46 0 0 ESP Atl\xe9tico Madrid
## 4 Pedri Normal LM 30 28 0 0 ESP Barcelona
## 5 Frenkie de Jong Normal CM 29 24 0 1 NED Barcelona
## Age Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY
## 1 26 1995 32 2529 28.1 11 10 11 0 0 6
## 2 26 1995 33 2022 22.5 7 8 7 0 0 3
## 3 29 1992 32 2635 29.3 1 2 1 0 0 8
## 4 18 2002 32 2130 23.7 2 3 2 0 0 2
## 5 23 1997 32 2721 30.2 3 4 3 0 0 4
## CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0 0.39 0.36 0.75 0.39 0.75
## 2 0 0.31 0.36 0.67 0.31 0.67
## 3 0 0.03 0.07 0.10 0.03 0.10
## 4 0 0.08 0.13 0.21 0.08 0.21
## 5 0 0.10 0.13 0.23 0.10 0.23
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 1 60 22 38 73 NA 0.2397710 0.7602290 TOTS
## 2 Matches 1 60 22 38 73 NA 0.4097860 0.5902140 TOTS
## 3 Matches 1 60 22 38 73 NA 0.4153839 0.5846161 TOTS
## 4 Matches 3 76 29 47 71 NA 0.4451463 0.5548537 TOTS
## 5 Matches 3 76 29 47 71 NA 0.4470152 0.5529848 TOTS## Player revision position Int TklW OG PKcon Nation Squad
## 1 Jordi Alba Normal LB 36 21 1 0 ESP Barcelona
## 2 Stefan Savic Normal CB 29 25 0 1 MNE Atl\xe9tico Madrid
## 3 Raphael Varane Normal CB 35 11 1 0 FRA Real Madrid
## 4 Mario Hermoso Normal CB 28 34 1 0 ESP Atl\xe9tico Madrid
## 5 Felipe Normal CB 37 15 1 0 BRA Atl\xe9tico Madrid
## Age Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY
## 1 32 1989 29 2530 28.1 3 5 3 0 0 8
## 2 30 1991 29 2593 28.8 1 0 1 0 0 13
## 3 28 1993 29 2580 28.7 2 0 2 0 0 2
## 4 25 1995 26 2181 24.2 1 1 1 0 0 5
## 5 31 1989 26 1688 18.8 0 0 0 0 0 6
## CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0 0.11 0.18 0.28 0.11 0.28
## 2 0 0.03 0.00 0.03 0.03 0.03
## 3 0 0.07 0.00 0.07 0.07 0.07
## 4 0 0.04 0.04 0.08 0.04 0.08
## 5 0 0.00 0.00 0.00 0.00 0.00
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 3 76 29 47 71 NA 0.4553549 0.5446451 TOTS
## 2 Matches 1 60 22 38 73 NA 0.4982691 0.5017309 TOTS
## 3 Matches 2 56 24 32 71 NA 0.5324940 0.4675060 Normal
## 4 Matches 1 60 22 38 73 NA 0.6077316 0.3922684 Normal
## 5 Matches 1 60 22 38 73 NA 0.6163003 0.3836997 Normal
French Ligue 1
ligue1_modeling <- ligue1_fifa17_modeling %>%
bind_rows(ligue1_fifa18_modeling, ligue1_fifa19_modeling) %>%
mutate(revision = as.factor(revision), Nation = as.factor(Nation))
ligue1_modeling %>%
ggplot(aes(x = minutes_played_divided_by90, fill = revision)) +
geom_density(alpha = 0.5)
ligue1_modeling_outfield <- ligue1_modeling %>%
filter(position != "GK") %>%
filter(minutes_played_divided_by90 >= 19) %>%
mutate(position = ifelse(position == "RWB", "RB", ifelse(position == "LWB", "LB", position))) %>%
select(-Goals_allowed, -GA90, -SoTA, -Saves, -Save_percent, -W, -L, -D, -CS, -CS_percent, -Pkatt_against, -PKA, -PKsv, -Pk_Save_percent, -PKm)set.seed(494)
ligue1_split <- initial_split(ligue1_modeling_outfield, prop = .75, strata = "revision")
ligue1_training <- training(ligue1_split)
ligue1_testing <- testing(ligue1_split)ligue1_ranger_recipe <- recipe(revision ~., data = ligue1_training) %>%
step_rm(Player, Nation, Squad, Born, minutes_played_divided_by90, G_per90, A_per90, Attendance) %>%
step_upsample(revision, over_ratio = .3) %>%
step_mutate_at(all_numeric(), fn = ~as.numeric(.))
ligue1_ranger_recipe %>%
prep(ligue1_training) %>%
juice()## # A tibble: 466 x 24
## position Int TklW OG PKcon Age MP Min Gls Ast Non_PK_G PK
## <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 RB 39 45 0 0 26 27 2149 0 3 0 0
## 2 LM 11 26 0 0 23 38 2476 3 5 3 0
## 3 RB 51 31 0 0 22 27 2324 0 1 0 0
## 4 ST 14 16 0 0 26 36 2225 10 4 10 0
## 5 CB 41 36 0 0 32 32 2635 1 0 1 0
## 6 RB 72 74 0 3 29 27 2395 0 1 0 0
## 7 RB 67 31 0 1 26 26 2198 1 1 1 0
## 8 CB 27 26 0 1 25 34 3015 1 0 1 0
## 9 RB 74 63 0 0 23 30 2646 0 4 0 0
## 10 LB 24 23 0 3 22 27 2121 0 4 0 0
## # … with 456 more rows, and 12 more variables: PKatt <dbl>, CrdY <dbl>,
## # CrdR <dbl>, G_plus_A_per90 <dbl>, G_minus_Pk_per90 <dbl>,
## # G_plus_A_minus_PK_per90 <dbl>, Rk <dbl>, GF <dbl>, GA <dbl>, GD <dbl>,
## # Pts <dbl>, revision <fct>ligue1_ranger <- rand_forest(mtry = tune(),
min_n = tune(),
trees = 100) %>%
set_mode("classification") %>%
set_engine("ranger")
ligue1_ranger_wf <-
workflow() %>%
add_recipe(ligue1_ranger_recipe) %>%
add_model(ligue1_ranger)
ligue1_ranger_wf## ══ Workflow ════════════════════════════════════════════════════════════════════
## Preprocessor: Recipe
## Model: rand_forest()
##
## ── Preprocessor ────────────────────────────────────────────────────────────────
## 3 Recipe Steps
##
## ● step_rm()
## ● step_upsample()
## ● step_mutate_at()
##
## ── Model ───────────────────────────────────────────────────────────────────────
## Random Forest Model Specification (classification)
##
## Main Arguments:
## mtry = tune()
## trees = 100
## min_n = tune()
##
## Computational engine: rangerset.seed(494)
ligue1_cv <- vfold_cv(ligue1_training, v = 5)
ligue1_rf_grid <- grid_regular(min_n(), finalize(mtry(), ligue1_training %>% select(-revision)), levels = 3)
ligue1_ctrl_res <- control_stack_grid()
ligue1_ranger_cv <- ligue1_ranger_wf %>%
tune_grid(resamples = ligue1_cv,
grid = ligue1_rf_grid,
control = ligue1_ctrl_res)## ! Fold1: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold1: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold1: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## # A tibble: 18 x 8
## mtry min_n .metric .estimator mean n std_err .config
## <int> <int> <chr> <chr> <dbl> <int> <dbl> <chr>
## 1 1 2 accuracy binary 0.912 5 0.0198 Preprocessor1_Model1
## 2 1 2 roc_auc binary 0.938 5 0.0238 Preprocessor1_Model1
## 3 1 21 accuracy binary 0.912 5 0.0213 Preprocessor1_Model2
## 4 1 21 roc_auc binary 0.942 5 0.0187 Preprocessor1_Model2
## 5 1 40 accuracy binary 0.912 5 0.0213 Preprocessor1_Model3
## 6 1 40 roc_auc binary 0.944 5 0.0176 Preprocessor1_Model3
## 7 16 2 accuracy binary 0.926 5 0.0219 Preprocessor1_Model4
## 8 16 2 roc_auc binary 0.943 5 0.0233 Preprocessor1_Model4
## 9 16 21 accuracy binary 0.912 5 0.0179 Preprocessor1_Model5
## 10 16 21 roc_auc binary 0.935 5 0.0259 Preprocessor1_Model5
## 11 16 40 accuracy binary 0.912 5 0.0183 Preprocessor1_Model6
## 12 16 40 roc_auc binary 0.924 5 0.0299 Preprocessor1_Model6
## 13 31 2 accuracy binary 0.924 5 0.0178 Preprocessor1_Model7
## 14 31 2 roc_auc binary 0.935 5 0.0259 Preprocessor1_Model7
## 15 31 21 accuracy binary 0.909 5 0.0192 Preprocessor1_Model8
## 16 31 21 roc_auc binary 0.929 5 0.0277 Preprocessor1_Model8
## 17 31 40 accuracy binary 0.907 5 0.0208 Preprocessor1_Model9
## 18 31 40 roc_auc binary 0.929 5 0.0263 Preprocessor1_Model9ligue1_best1 <- ligue1_ranger_cv %>%
select_best(metric = "accuracy")
ligue1_ranger_final_wf<- ligue1_ranger_wf %>%
finalize_workflow(ligue1_best1)ligue1_ranger_fit <- ligue1_ranger_final_wf %>%
fit(ligue1_training)
ligue1_rf_explain <-
explain_tidymodels(
model = ligue1_ranger_fit,
data = ligue1_training %>% select(-revision),
y = as.numeric(ligue1_training$revision == "TOTS"),
label = "rf"
)## Preparation of a new explainer is initiated
## -> model label : rf
## -> data : 407 rows 31 cols
## -> target variable : 407 values
## -> predict function : yhat.workflow will be used ( [33m default [39m )
## -> predicted values : No value for predict function target column. ( [33m default [39m )
## -> model_info : package tidymodels , ver. 0.1.3 , task classification ( [33m default [39m )
## -> predicted values : numerical, min = 0 , mean = 0.147113 , max = 1
## -> residual function : difference between y and yhat ( [33m default [39m )
## -> residuals : numerical, min = -0.86 , mean = -0.0291769 , max = 0.53
## [32m A new explainer has been created! [39m
ligue1_ranger_test <- ligue1_ranger_final_wf %>%
last_fit(ligue1_split)
ligue1_ranger_test %>%
collect_metrics()## # A tibble: 2 x 4
## .metric .estimator .estimate .config
## <chr> <chr> <dbl> <chr>
## 1 accuracy binary 0.873 Preprocessor1_Model1
## 2 roc_auc binary 0.924 Preprocessor1_Model1ligue1_preds1 <- ligue1_ranger_test %>%
collect_predictions()
ligue1_preds1 %>%
conf_mat(revision, .pred_class)## Truth
## Prediction Normal TOTS
## Normal 108 6
## TOTS 11 9ligue1_ranger_test <- ligue1_testing %>%
bind_cols(predict(ligue1_ranger_fit, new_data = ligue1_testing, type = "prob")) %>%
bind_cols(predict(ligue1_ranger_fit, new_data = ligue1_testing)) ## Truth
## Prediction Normal TOTS
## Normal 108 5
## TOTS 11 10## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Lois Diony TOTS ST 4 14 0 0 FRA Dijon 23
## 2 Blaise Matuidi Normal CDM 40 42 0 0 FRA Paris S-G 29
## 3 Djibril Sidibe Normal RB 47 52 0 1 FRA Monaco 24
## 4 Jemerson Normal CB 54 51 0 0 BRA Monaco 23
## 5 Giovani Lo Celso Normal CAM 20 59 0 0 ARG Paris S-G 21
## 6 Dimitri Payet Normal LW 18 9 0 0 FRA Marseille 30
## 7 Alassane Plea Normal ST 13 9 0 0 FRA Nice 24
## 8 Adil Rami TOTS CB 33 20 1 0 FRA Marseille 31
## 9 Dani Alves Normal RB 28 52 0 0 BRA Paris S-G 34
## 10 Jorge Normal LB 52 33 0 0 BRA Monaco 21
## 11 Radamel Falcao TOTS ST 13 7 1 0 COL Monaco 31
## 12 Joao Moutinho Normal CM 39 44 0 0 POR Monaco 30
## 13 Houssem Aouar Normal CM 31 36 0 0 FRA Lyon 20
## 14 Kenny Lala TOTS RB 29 43 0 1 FRA Strasbourg 26
## 15 Ferland Mendy TOTS LB 25 30 0 1 FRA Lyon 23
## 16 Zeki Celik Normal RB 34 55 0 3 TUR Lille 21
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1992 35 2807 31.2 11 7 11 0 0 2 1
## 2 1987 34 2415 26.8 4 4 4 0 0 4 0
## 3 1992 29 2321 25.8 2 5 2 0 0 7 0
## 4 1992 34 3058 34.0 2 0 2 0 0 8 2
## 5 1996 33 1776 19.7 4 2 4 0 0 2 0
## 6 1987 31 2347 26.1 6 13 4 2 3 6 0
## 7 1993 35 3041 33.8 16 4 15 1 2 7 0
## 8 1985 33 2955 32.8 1 1 1 0 0 5 0
## 9 1983 25 2065 22.9 1 4 1 0 0 7 1
## 10 1996 22 1919 21.3 1 2 1 0 0 8 0
## 11 1986 26 2128 23.6 18 2 15 3 4 1 0
## 12 1986 33 2802 31.1 1 4 1 0 0 6 0
## 13 1998 37 3061 34.0 7 7 7 0 0 2 0
## 14 1991 34 3060 34.0 5 9 4 1 2 4 0
## 15 1995 30 2531 28.1 2 1 2 0 0 2 0
## 16 1997 34 2971 33.0 1 5 1 0 0 5 1
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90 Rk
## 1 0.35 0.22 0.58 0.35 0.58 16
## 2 0.15 0.15 0.30 0.15 0.30 2
## 3 0.08 0.19 0.27 0.08 0.27 1
## 4 0.06 0.00 0.06 0.06 0.06 1
## 5 0.20 0.10 0.30 0.20 0.30 1
## 6 0.23 0.50 0.73 0.15 0.65 4
## 7 0.47 0.12 0.59 0.44 0.56 8
## 8 0.03 0.03 0.06 0.03 0.06 4
## 9 0.04 0.17 0.22 0.04 0.22 1
## 10 0.05 0.09 0.14 0.05 0.14 2
## 11 0.76 0.08 0.85 0.63 0.72 2
## 12 0.03 0.13 0.16 0.03 0.16 2
## 13 0.21 0.21 0.41 0.21 0.41 3
## 14 0.15 0.26 0.41 0.12 0.38 11
## 15 0.07 0.04 0.11 0.07 0.11 3
## 16 0.03 0.15 0.18 0.03 0.18 2
## GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 46 58 -12 37 10126 0.720 0.280 Normal
## 2 83 27 56 87 45160 0.360 0.640 TOTS
## 3 107 31 76 95 9586 0.180 0.820 TOTS
## 4 107 31 76 95 9586 0.130 0.870 TOTS
## 5 108 29 79 93 46929 0.375 0.625 TOTS
## 6 80 47 33 77 46040 0.460 0.540 TOTS
## 7 53 52 1 54 22876 0.270 0.730 TOTS
## 8 80 47 33 77 46040 0.740 0.260 Normal
## 9 108 29 79 93 46929 0.170 0.830 TOTS
## 10 85 45 40 80 9243 0.475 0.525 TOTS
## 11 85 45 40 80 9243 0.840 0.160 Normal
## 12 85 45 40 80 9243 0.200 0.800 TOTS
## 13 70 47 23 72 49079 0.290 0.710 TOTS
## 14 58 48 10 49 25216 0.790 0.210 Normal
## 15 70 47 23 72 49079 0.720 0.280 Normal
## 16 68 33 35 75 34079 0.365 0.635 TOTSligue1_modeling21 <- fifa21_modeling_ligue1 %>%
mutate(revision = as.factor(revision), Nation = as.factor(Nation), Age = as.integer(Age))ligue1_modeling_outfield21 <- ligue1_modeling21 %>%
filter(position != "GK") %>%
#filter(position %in% c("ST", "LW", "RW", "CF", "CAM")) %>%
filter(minutes_played_divided_by90 >= 14) %>%
mutate(position = ifelse(position == "RWB", "RB", ifelse(position == "LWB", "LB", position)))ligue1_ranger_test21 <- ligue1_modeling_outfield21 %>%
bind_cols(predict(ligue1_ranger_fit, new_data = ligue1_modeling_outfield21, type = "prob")) %>%
bind_cols(predict(ligue1_ranger_fit, new_data = ligue1_modeling_outfield21)) ## Warning: Novel levels found in column 'Nation': 'AUT', 'CAN', 'CHI', 'CRC',
## 'ECU', 'PER', 'SCO', 'ZIM'. The levels have been removed, and values have been
## coerced to 'NA'.
## Warning: Novel levels found in column 'Nation': 'AUT', 'CAN', 'CHI', 'CRC',
## 'ECU', 'PER', 'SCO', 'ZIM'. The levels have been removed, and values have been
## coerced to 'NA'.ligue1_ranger_test21 %>%
filter(position %in% c("ST", "RW", "CF", "LW")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Memphis Depay Normal CF 9 5 0 0 NED Lyon 27
## 2 Gaetan Laborde Normal ST 14 30 0 0 FRA Montpellier 26
## 3 Kylian Mbappe Normal ST 5 4 0 0 FRA Paris S-G 22
## 4 Kevin Volland Normal ST 3 26 0 0 GER Monaco 28
## 5 Wissam Ben Yedder Normal ST 8 13 0 0 FRA Monaco 30
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1994 34 2653 29.5 18 9 10 8 8 4 0
## 2 1994 34 2932 32.6 13 8 13 0 0 3 0
## 3 1998 29 2214 24.6 25 7 19 6 6 5 0
## 4 1992 31 2419 26.9 15 7 15 0 0 3 0
## 5 1990 33 2266 25.2 18 5 9 9 11 3 0
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.61 0.31 0.92 0.34 0.64
## 2 0.40 0.25 0.64 0.40 0.64
## 3 1.02 0.28 1.30 0.77 1.06
## 4 0.56 0.26 0.82 0.56 0.82
## 5 0.71 0.20 0.91 0.36 0.56
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 4 67 35 32 67 258 0.17 0.83 TOTS
## 2 Matches 8 54 57 -3 47 NA 0.20 0.80 TOTS
## 3 Matches 2 77 26 51 72 NA 0.38 0.62 TOTS
## 4 Matches 3 71 38 33 71 NA 0.38 0.62 TOTS
## 5 Matches 3 71 38 33 71 NA 0.41 0.59 TOTSligue1_ranger_test21 %>%
filter(position %in% c("CAM", "CM", "CDM", "LM", "RM")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Jonathan Bamba Normal LM 27 31 0 0 FRA Lille 25
## 2 Aurelien Tchouameni Normal CM 55 86 0 0 FRA Monaco 21
## 3 Ander Herrera Normal CM 22 32 0 0 ESP Paris S-G 31
## 4 Gael Kakuta Normal CAM 21 26 0 1 COD Lens 29
## 5 Leandro Paredes Normal CM 16 25 0 0 ARG Paris S-G 26
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1996 34 2719 30.2 6 9 6 0 0 2 0
## 2 2000 32 2703 30.0 2 4 2 0 0 9 1
## 3 1989 27 1571 17.5 1 3 1 0 0 3 0
## 4 1991 31 2217 24.6 11 5 4 7 9 2 0
## 5 1994 20 1288 14.3 1 2 1 0 0 7 1
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.20 0.30 0.50 0.20 0.50
## 2 0.07 0.13 0.20 0.07 0.20
## 3 0.06 0.17 0.23 0.06 0.23
## 4 0.45 0.20 0.65 0.16 0.37
## 5 0.07 0.14 0.21 0.07 0.21
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 1 57 22 35 73 234 0.390 0.610 TOTS
## 2 Matches 3 71 38 33 71 NA 0.470 0.530 TOTS
## 3 Matches 2 77 26 51 72 NA 0.645 0.355 Normal
## 4 Matches 5 54 46 8 56 NA 0.660 0.340 Normal
## 5 Matches 2 77 26 51 72 NA 0.670 0.330 Normalligue1_ranger_test21 %>%
filter(position %in% c("LB", "CB", "RB")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Thomas Delaine Normal LB 17 15 0 0 FRA Metz 29
## 2 Leonardo Balerdi Normal CB 29 17 0 0 ARG Marseille 22
## 3 Leo Dubois Normal RB 22 29 0 1 FRA Lyon 26
## 4 Senou Coulibaly Normal CB 34 20 0 0 MLI Dijon 26
## 5 Sven Botman Normal CB 36 27 0 0 NED Lille 21
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1992 22 1600 17.8 3 1 3 0 0 4 0
## 2 1999 17 1363 15.1 2 0 2 0 0 5 1
## 3 1994 33 2610 29.0 2 3 2 0 0 3 0
## 4 1994 19 1664 18.5 2 0 2 0 0 6 1
## 5 2000 33 2949 32.8 0 0 0 0 0 2 0
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.17 0.06 0.23 0.17 0.23
## 2 0.13 0.00 0.13 0.13 0.13
## 3 0.07 0.10 0.17 0.07 0.17
## 4 0.11 0.00 0.11 0.11 0.11
## 5 0.00 0.00 0.00 0.00 0.00
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 10 37 41 -4 43 557 0.560 0.440 Normal
## 2 Matches 6 49 42 7 55 NA 0.585 0.415 Normal
## 3 Matches 4 67 35 32 67 258 0.630 0.370 Normal
## 4 Matches 20 23 61 -38 18 NA 0.645 0.355 Normal
## 5 Matches 1 57 22 35 73 234 0.665 0.335 Normal
German Bundesliga
fifa19_modeling_bundesliga2 <- fifa19_modeling_bundesliga %>%
mutate(Player = paste(Player, '19'))
fifa18_modeling_bundesliga2 <- fifa18_modeling_bundesliga %>%
mutate(Player = paste(Player, '18'))
fifa17_modeling_bundesliga2 <- fifa17_modeling_bundesliga %>%
mutate(Player = paste(Player, '17'))
bundesliga_modeling <- fifa17_modeling_bundesliga2%>%
bind_rows(fifa18_modeling_bundesliga2, fifa19_modeling_bundesliga2) %>%
mutate(revision = as.factor(revision), Nation = as.factor(Nation))bundesliga_modeling %>%
select(where(is.numeric)) %>%
pivot_longer(cols = everything(),
names_to = "variable",
values_to = "value") %>%
ggplot(aes(x = value)) +
geom_histogram(bins = 30) +
facet_wrap(vars(variable),
scales = "free")## Warning: Removed 17990 rows containing non-finite values (stat_bin).
bundesliga_modeling %>%
ggplot(aes(x = revision, fill = revision)) +
geom_bar() +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold"))
bundesliga_modeling %>%
ggplot(aes(x = Gls, fill = revision)) +
geom_density(alpha = 0.5) +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold")) +
xlab("Goals")
bundesliga_modeling %>%
ggplot(aes(x = Rk, fill = revision)) +
geom_density(alpha = 0.5) +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold")) +
xlab("Table Position")
bundesliga_modeling %>%
ggplot(aes(x = minutes_played_divided_by90, fill = revision)) +
geom_density(alpha = 0.5) +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold")) +
xlab("Total Minutes Played Dived by 90 (Full Games Played)")
bundesliga_modeling %>%
ggplot(aes(x = position, fill = revision)) +
geom_bar(position = "dodge") +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold"))
bundesliga_modeling_outfield <- bundesliga_modeling %>%
filter(position != "GK") %>%
#filter(position %in% c("ST", "LW", "RW", "CF", "CAM")) %>%
filter(minutes_played_divided_by90 >= 19) %>%
mutate(position = ifelse(position == "RWB", "RB", ifelse(position == "LWB", "LB", ifelse(position == "RW", "RM", ifelse(position == "LW", "LM", position))))) %>%
select(-Goals_allowed, -GA90, -SoTA, -Saves, -Save_percent, -W, -L, -D, -CS, -CS_percent, -Pkatt_against, -PKA, -PKsv, -Pk_Save_percent, -PKm)set.seed(494)
bundesliga_split <- initial_split(bundesliga_modeling_outfield, prop = .75, strata = "revision")
bundesliga_training <- training(bundesliga_split)
bundesliga_testing <- testing(bundesliga_split)bundesliga_ranger_recipe <- recipe(revision ~., data = bundesliga_training) %>%
step_rm(Player, Nation, Squad, G_per90, A_per90, minutes_played_divided_by90, Attendance, Born) %>%
step_upsample(revision, over_ratio = .55) %>%
step_mutate_at(all_numeric(), fn = ~as.numeric(.))
bundesliga_ranger_recipe %>%
prep(bundesliga_training) %>%
juice()## # A tibble: 434 x 24
## position Int TklW OG PKcon Age MP Min Gls Ast Non_PK_G PK
## <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 CM 37 51 0 1 28 27 2302 1 3 1 0
## 2 LM 29 10 0 0 30 26 1982 7 4 6 1
## 3 ST 11 20 0 0 23 28 1735 5 1 5 0
## 4 CB 34 41 0 0 28 27 2315 3 2 3 0
## 5 CB 74 39 0 0 19 25 2106 0 0 0 0
## 6 CB 10 30 0 0 24 26 2126 3 0 3 0
## 7 LM 27 17 0 0 26 25 1729 1 3 1 0
## 8 CB 16 21 0 1 21 28 2427 0 2 0 0
## 9 CB 35 23 0 1 32 30 2637 0 0 0 0
## 10 CB 21 12 0 1 22 24 1901 1 0 1 0
## # … with 424 more rows, and 12 more variables: PKatt <dbl>, CrdY <dbl>,
## # CrdR <dbl>, G_plus_A_per90 <dbl>, G_minus_Pk_per90 <dbl>,
## # G_plus_A_minus_PK_per90 <dbl>, Rk <dbl>, GF <dbl>, GA <dbl>, GD <dbl>,
## # Pts <dbl>, revision <fct>bundesliga_ranger <- rand_forest(mtry = tune(),
min_n = tune(),
trees = 100) %>%
set_mode("classification") %>%
set_engine("ranger")
bundesliga_ranger_wf <-
workflow() %>%
add_recipe(bundesliga_ranger_recipe) %>%
add_model(bundesliga_ranger)
bundesliga_ranger_wf## ══ Workflow ════════════════════════════════════════════════════════════════════
## Preprocessor: Recipe
## Model: rand_forest()
##
## ── Preprocessor ────────────────────────────────────────────────────────────────
## 3 Recipe Steps
##
## ● step_rm()
## ● step_upsample()
## ● step_mutate_at()
##
## ── Model ───────────────────────────────────────────────────────────────────────
## Random Forest Model Specification (classification)
##
## Main Arguments:
## mtry = tune()
## trees = 100
## min_n = tune()
##
## Computational engine: rangerset.seed(494)
bundesliga_cv <- vfold_cv(bundesliga_training, v = 5)
bundesliga_rf_grid <- grid_regular(min_n(), finalize(mtry(), bundesliga_training %>% select(-revision)), levels = 3)
ctrl_res <- control_stack_grid()
bundesliga_ranger_cv <- bundesliga_ranger_wf %>%
tune_grid(resamples = bundesliga_cv,
grid = bundesliga_rf_grid,
control = ctrl_res)## ! Fold1: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold1: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold1: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## # A tibble: 18 x 8
## mtry min_n .metric .estimator mean n std_err .config
## <int> <int> <chr> <chr> <dbl> <int> <dbl> <chr>
## 1 1 2 accuracy binary 0.884 5 0.0130 Preprocessor1_Model1
## 2 1 2 roc_auc binary 0.907 5 0.0111 Preprocessor1_Model1
## 3 1 21 accuracy binary 0.893 5 0.0133 Preprocessor1_Model2
## 4 1 21 roc_auc binary 0.890 5 0.0160 Preprocessor1_Model2
## 5 1 40 accuracy binary 0.884 5 0.0146 Preprocessor1_Model3
## 6 1 40 roc_auc binary 0.900 5 0.0105 Preprocessor1_Model3
## 7 16 2 accuracy binary 0.869 5 0.0121 Preprocessor1_Model4
## 8 16 2 roc_auc binary 0.867 5 0.0202 Preprocessor1_Model4
## 9 16 21 accuracy binary 0.884 5 0.0146 Preprocessor1_Model5
## 10 16 21 roc_auc binary 0.862 5 0.0172 Preprocessor1_Model5
## 11 16 40 accuracy binary 0.875 5 0.0144 Preprocessor1_Model6
## 12 16 40 roc_auc binary 0.862 5 0.0169 Preprocessor1_Model6
## 13 31 2 accuracy binary 0.863 5 0.0196 Preprocessor1_Model7
## 14 31 2 roc_auc binary 0.864 5 0.0240 Preprocessor1_Model7
## 15 31 21 accuracy binary 0.863 5 0.0212 Preprocessor1_Model8
## 16 31 21 roc_auc binary 0.851 5 0.0140 Preprocessor1_Model8
## 17 31 40 accuracy binary 0.869 5 0.0181 Preprocessor1_Model9
## 18 31 40 roc_auc binary 0.852 5 0.0175 Preprocessor1_Model9bundesliga_best1 <- bundesliga_ranger_cv %>%
select_best(metric = "accuracy")
bundesliga_ranger_final_wf<- bundesliga_ranger_wf %>%
finalize_workflow(bundesliga_best1)set.seed(494)
bundesliga_ranger_fit <- bundesliga_ranger_final_wf %>%
fit(bundesliga_training)
bundesliga_rf_explain <-
explain_tidymodels(
model = bundesliga_ranger_fit,
data = bundesliga_training %>% select(-revision),
y = as.numeric(bundesliga_training$revision == "TOTS"),
label = "rf"
)## Preparation of a new explainer is initiated
## -> model label : rf
## -> data : 328 rows 31 cols
## -> target variable : 328 values
## -> predict function : yhat.workflow will be used ( [33m default [39m )
## -> predicted values : No value for predict function target column. ( [33m default [39m )
## -> model_info : package tidymodels , ver. 0.1.3 , task classification ( [33m default [39m )
## -> predicted values : numerical, min = 0.01184855 , mean = 0.235783 , max = 0.9191418
## -> residual function : difference between y and yhat ( [33m default [39m )
## -> residuals : numerical, min = -0.7846212 , mean = -0.08944149 , max = 0.714248
## [32m A new explainer has been created! [39m
bundesliga_ranger_test <- bundesliga_ranger_final_wf %>%
last_fit(bundesliga_split)
bundesliga_ranger_test %>%
collect_metrics()## # A tibble: 2 x 4
## .metric .estimator .estimate .config
## <chr> <chr> <dbl> <chr>
## 1 accuracy binary 0.835 Preprocessor1_Model1
## 2 roc_auc binary 0.862 Preprocessor1_Model1preds1 <- bundesliga_ranger_test %>%
collect_predictions()
preds1 %>%
conf_mat(revision, .pred_class)## Truth
## Prediction Normal TOTS
## Normal 84 9
## TOTS 9 7bundesliga_ranger_test <- bundesliga_testing %>%
bind_cols(predict(bundesliga_ranger_fit, new_data = bundesliga_testing, type = "prob")) %>%
bind_cols(predict(bundesliga_ranger_fit, new_data = bundesliga_testing)) ## Truth
## Prediction Normal TOTS
## Normal 85 9
## TOTS 8 7## Player revision position Int TklW OG PKcon Nation
## 1 Kerem Demirbay 17 Normal CAM 44 33 0 0 GER
## 2 Marco Fabian 17 TOTS CAM 51 31 0 0 MEX
## 3 Vincenzo Grifo 17 TOTS LM 37 31 0 0 ITA
## 4 Sebastian Rudy 17 TOTS CM 99 64 0 0 GER
## 5 Javi Martinez 17 Normal CB 55 37 0 0 ESP
## 6 Julian Brandt 18 Normal LM 15 16 0 0 GER
## 7 Michael Gregoritsch 18 TOTS CAM 12 15 0 0 AUT
## 8 Thorgan Hazard 18 TOTS LM 20 33 0 0 BEL
## 9 Naby Keita 18 TOTS CM 22 33 0 0 GUI
## 10 Andrej Kramaric 18 Normal ST 8 3 0 0 CRO
## 11 Philipp Max 18 TOTS LB 19 31 0 0 GER
## 12 Nils Petersen 18 TOTS ST 15 16 0 0 GER
## 13 Wendell 18 TOTS LB 20 25 0 0 BRA
## 14 Ishak Belfodil 19 Normal ST 2 9 0 0 ALG
## 15 Mats Hummels 19 Normal CB 27 13 0 1 GER
## 16 Andrej Kramaric 19 Normal ST 8 11 0 0 CRO
## 17 Lukasz Piszczek 19 Normal RB 31 30 0 0 POL
## Squad Age Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G
## 1 Hoffenheim 23 1993 28 2169 24.1 6 8 6
## 2 Eint Frankfurt 27 1989 24 2054 22.8 7 4 6
## 3 Freiburg 23 1993 30 2492 27.7 6 7 5
## 4 Hoffenheim 26 1990 32 2786 31.0 2 6 2
## 5 Bayern Munich 27 1988 25 2131 23.7 1 1 1
## 6 Leverkusen 21 1996 34 2326 25.8 9 3 9
## 7 Augsburg 23 1994 32 2527 28.1 13 3 12
## 8 M'Gladbach 24 1993 34 2939 32.7 10 5 5
## 9 RB Leipzig 22 1995 27 1962 21.8 6 5 6
## 10 Hoffenheim 26 1991 34 2228 24.8 13 6 11
## 11 Augsburg 23 1993 33 2959 32.9 2 12 2
## 12 Freiburg 28 1988 32 2244 24.9 15 1 10
## 13 Leverkusen 24 1993 26 2115 23.5 2 3 0
## 14 Hoffenheim 26 1992 28 1863 20.7 16 3 16
## 15 Bayern Munich 29 1988 21 1775 19.7 1 1 1
## 16 Hoffenheim 27 1991 30 2396 26.6 17 4 12
## 17 Dortmund 33 1985 20 1756 19.5 1 6 1
## PK PKatt CrdY CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90
## 1 0 0 4 0 0.25 0.33 0.58 0.25
## 2 1 2 10 0 0.31 0.18 0.48 0.26
## 3 1 1 1 0 0.22 0.25 0.47 0.18
## 4 0 0 9 0 0.06 0.19 0.26 0.06
## 5 0 0 5 0 0.04 0.04 0.08 0.04
## 6 0 0 0 0 0.35 0.12 0.46 0.35
## 7 1 1 3 0 0.46 0.11 0.57 0.43
## 8 5 6 1 0 0.31 0.15 0.46 0.15
## 9 0 0 8 2 0.28 0.23 0.50 0.28
## 10 2 2 1 0 0.53 0.24 0.77 0.44
## 11 0 0 5 0 0.06 0.36 0.43 0.06
## 12 5 6 4 1 0.60 0.04 0.64 0.40
## 13 2 3 7 1 0.09 0.13 0.21 0.00
## 14 0 0 3 0 0.77 0.14 0.92 0.77
## 15 0 0 1 0 0.05 0.05 0.10 0.05
## 16 5 6 2 0 0.64 0.15 0.79 0.45
## 17 0 0 3 0 0.05 0.31 0.36 0.05
## G_plus_A_minus_PK_per90 Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS
## 1 0.58 4 64 37 27 62 28155 0.4478120 0.5521880
## 2 0.44 11 36 43 -7 42 49165 0.8021510 0.1978490
## 3 0.43 7 42 60 -18 48 23959 0.8394151 0.1605849
## 4 0.26 4 64 37 27 62 28155 0.5792912 0.4207088
## 5 0.08 1 89 22 67 82 75000 0.2754054 0.7245946
## 6 0.46 5 58 44 14 55 28415 0.4763345 0.5236655
## 7 0.53 12 43 46 -3 41 28238 0.7306608 0.2693392
## 8 0.31 9 47 52 -5 47 50986 0.6338276 0.3661724
## 9 0.50 6 57 53 4 53 39397 0.7875857 0.2124143
## 10 0.69 3 66 48 18 55 28716 0.2964712 0.7035288
## 11 0.43 12 43 46 -3 41 28238 0.5823812 0.4176188
## 12 0.44 15 32 56 -24 36 23894 0.7362705 0.2637295
## 13 0.13 5 58 44 14 55 28415 0.7559639 0.2440361
## 14 0.92 9 70 52 18 51 28456 0.3138931 0.6861069
## 15 0.10 1 88 32 56 78 75000 0.3905305 0.6094695
## 16 0.60 9 70 52 18 51 28456 0.3151089 0.6848911
## 17 0.36 2 81 44 37 76 80841 0.4234112 0.5765888
## .pred_class
## 1 TOTS
## 2 Normal
## 3 Normal
## 4 Normal
## 5 TOTS
## 6 TOTS
## 7 Normal
## 8 Normal
## 9 Normal
## 10 TOTS
## 11 Normal
## 12 Normal
## 13 Normal
## 14 TOTS
## 15 TOTS
## 16 TOTS
## 17 TOTSbundesliga_modeling21 <- fifa21_modeling_bundesliga %>%
mutate(revision = as.factor(revision), Nation = as.factor(Nation), Age = as.integer(Age))bundesliga_modeling_outfield21 <- bundesliga_modeling21 %>%
filter(position != "GK") %>%
#filter(position %in% c("ST", "LW", "RW", "CF", "CAM")) %>%
filter(minutes_played_divided_by90 >= 18) %>%
mutate(position = ifelse(position == "RWB", "RB", ifelse(position == "LWB", "LB", ifelse(position == "RW", "RM", ifelse(position == "LW", "LM", position)))))bundesliga_ranger_test21 <- bundesliga_modeling_outfield21 %>%
bind_cols(predict(bundesliga_ranger_fit, new_data = bundesliga_modeling_outfield21, type = "prob")) %>%
bind_cols(predict(bundesliga_ranger_fit, new_data = bundesliga_modeling_outfield21)) ## Warning: Novel levels found in column 'Nation': 'ANG', 'ARM', 'BEN', 'BFA',
## 'BUL', 'CAN', 'ECU', 'FRO', 'MKD', 'WAL'. The levels have been removed, and
## values have been coerced to 'NA'.
## Warning: Novel levels found in column 'Nation': 'ANG', 'ARM', 'BEN', 'BFA',
## 'BUL', 'CAN', 'ECU', 'FRO', 'MKD', 'WAL'. The levels have been removed, and
## values have been coerced to 'NA'.bundesliga_ranger_test21 %>%
filter(position %in% c("ST", "RW", "CF", "LW")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad
## 1 Wout Weghorst Normal ST 6 11 0 0 NED Wolfsburg
## 2 Robert Lewandowski Normal ST 6 12 0 0 POL Bayern Munich
## 3 Erling Haaland Normal ST 4 6 0 0 NOR Dortmund
## 4 Andre Silva Normal ST 4 4 0 0 POR Eint Frankfurt
## 5 Sasa Kalajdzic Normal ST 5 6 0 0 AUT Stuttgart
## Age Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY
## 1 28 1992 31 2671 29.7 20 7 18 2 3 3
## 2 32 1988 26 2188 24.3 36 6 30 6 7 4
## 3 20 2000 26 2227 24.7 25 5 23 2 4 2
## 4 25 1995 29 2490 27.7 25 6 19 6 6 1
## 5 23 1997 30 1874 20.8 14 4 14 0 0 1
## CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0 0.67 0.24 0.91 0.61 0.84
## 2 0 1.48 0.25 1.73 1.23 1.48
## 3 0 1.01 0.20 1.21 0.93 1.13
## 4 0 0.90 0.22 1.12 0.69 0.90
## 5 0 0.67 0.19 0.86 0.67 0.86
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 3 54 32 22 57 610 0.1393530 0.8606470 TOTS
## 2 Matches 1 86 40 46 71 NA 0.2242189 0.7757811 TOTS
## 3 Matches 5 66 42 24 55 1407 0.2291923 0.7708077 TOTS
## 4 Matches 4 62 47 15 56 967 0.2335847 0.7664153 TOTS
## 5 Matches 10 52 51 1 39 1108 0.4769445 0.5230555 TOTSbundesliga_ranger_test21 %>%
filter(position %in% c("CAM", "CM", "CDM", "LM", "RM")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Thomas Muller Normal CAM 17 32 0 0 GER Bayern Munich 31
## 2 Joshua Kimmich Normal CDM 36 29 0 0 GER Bayern Munich 26
## 3 Leroy Sane Normal LM 11 19 0 0 GER Bayern Munich 25
## 4 Marcel Sabitzer Normal CM 34 25 0 0 AUT RB Leipzig 27
## 5 Leon Goretzka Normal CM 49 27 0 0 GER Bayern Munich 26
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1989 29 2453 27.3 10 17 9 1 1 0 0
## 2 1995 24 1924 21.4 3 10 3 0 0 4 0
## 3 1996 29 1672 18.6 4 9 4 0 0 2 0
## 4 1994 24 1756 19.5 7 2 4 3 3 6 0
## 5 1995 23 1695 18.8 5 5 5 0 0 2 0
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.37 0.62 0.99 0.33 0.95
## 2 0.14 0.47 0.61 0.14 0.61
## 3 0.22 0.48 0.70 0.22 0.70
## 4 0.36 0.10 0.46 0.21 0.31
## 5 0.27 0.27 0.53 0.27 0.53
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 1 86 40 46 71 NA 0.2481039 0.7518961 TOTS
## 2 Matches 1 86 40 46 71 NA 0.3504694 0.6495306 TOTS
## 3 Matches 1 86 40 46 71 NA 0.3699144 0.6300856 TOTS
## 4 Matches 2 55 25 30 64 1125 0.3781275 0.6218725 TOTS
## 5 Matches 1 86 40 46 71 NA 0.3908449 0.6091551 TOTSbundesliga_ranger_test21 %>%
filter(position %in% c("LB", "CB", "RB")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 David Alaba Normal CB 34 28 0 0 AUT Bayern Munich 28
## 2 Jerome Boateng Normal CB 42 17 0 0 GER Bayern Munich 32
## 3 Willi Orban Normal CB 29 22 0 0 HUN RB Leipzig 28
## 4 Ridle Baku Normal RB 43 26 0 0 GER Wolfsburg 23
## 5 Angelino Normal LB 29 16 0 0 ESP RB Leipzig 24
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1992 29 2454 27.3 2 2 2 0 0 3 0
## 2 1988 26 2148 23.9 1 1 1 0 0 6 0
## 3 1992 26 2093 23.3 4 1 4 0 0 4 0
## 4 1998 29 2409 26.8 6 4 6 0 0 0 0
## 5 1997 24 2042 22.7 4 4 4 0 0 2 0
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.07 0.07 0.15 0.07 0.15
## 2 0.04 0.04 0.08 0.04 0.08
## 3 0.17 0.04 0.22 0.17 0.22
## 4 0.22 0.15 0.37 0.22 0.37
## 5 0.18 0.18 0.35 0.18 0.35
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 1 86 40 46 71 NA 0.4763832 0.5236168 TOTS
## 2 Matches 1 86 40 46 71 NA 0.4878442 0.5121558 TOTS
## 3 Matches 2 55 25 30 64 1125 0.5158685 0.4841315 Normal
## 4 Matches 3 54 32 22 57 610 0.5356797 0.4643203 Normal
## 5 Matches 2 55 25 30 64 1125 0.5560889 0.4439111 Normal
Italian Serie A
fifa19_modeling_serie_a2 <- fifa19_modeling_serie_a %>%
mutate(Player = paste(Player, '19'))
fifa18_modeling_serie_a2 <- fifa18_modeling_serie_a %>%
mutate(Player = paste(Player, '18'))
fifa17_modeling_serie_a2 <- fifa17_modeling_serie_a %>%
mutate(Player = paste(Player, '17'))
serie_a_modeling <- fifa17_modeling_serie_a2%>%
bind_rows(fifa18_modeling_serie_a2, fifa19_modeling_serie_a2) %>%
mutate(revision = as.factor(revision), Nation = as.factor(Nation))serie_a_modeling %>%
select(where(is.numeric)) %>%
pivot_longer(cols = everything(),
names_to = "variable",
values_to = "value") %>%
ggplot(aes(x = value)) +
geom_histogram(bins = 30) +
facet_wrap(vars(variable),
scales = "free")## Warning: Removed 19578 rows containing non-finite values (stat_bin).
serie_a_modeling %>%
ggplot(aes(x = revision, fill = revision)) +
geom_bar() +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold"))
serie_a_modeling %>%
ggplot(aes(x = Gls, fill = revision)) +
geom_density(alpha = 0.5) +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold")) +
xlab("Goals")
serie_a_modeling %>%
ggplot(aes(x = Rk, fill = revision)) +
geom_density(alpha = 0.5) +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold")) +
xlab("Table Position")
serie_a_modeling %>%
ggplot(aes(x = minutes_played_divided_by90, fill = revision)) +
geom_density(alpha = 0.5) +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold")) +
xlab("Total Minutes Played Dived by 90 (Full Games Played)")
serie_a_modeling %>%
ggplot(aes(x = position, fill = revision)) +
geom_bar(position = "dodge") +
scale_fill_manual(values = c("TOTS" = "blue", "Normal" = "gold"))
serie_a_modeling_outfield <- serie_a_modeling %>%
filter(position != "GK") %>%
#filter(position %in% c("ST", "LW", "RW", "CF", "CAM")) %>%
filter(minutes_played_divided_by90 >= 19) %>%
mutate(position = ifelse(position == "RWB", "RB", ifelse(position == "LWB", "LB", ifelse(position == "RW", "RM", ifelse(position == "LW", "LM", position))))) %>%
select(-Goals_allowed, -GA90, -SoTA, -Saves, -Save_percent, -W, -L, -D, -CS, -CS_percent, -Pkatt_against, -PKA, -PKsv, -Pk_Save_percent, -PKm)set.seed(494)
serie_a_split <- initial_split(serie_a_modeling_outfield, prop = .75, strata = "revision")
serie_a_training <- training(serie_a_split)
serie_a_testing <- testing(serie_a_split)serie_a_ranger_recipe <- recipe(revision ~., data = serie_a_training) %>%
step_rm(Player, Nation, Squad, Born, G_per90, A_per90, minutes_played_divided_by90, Attendance) %>%
step_upsample(revision, over_ratio = .4) %>%
step_mutate_at(all_numeric(), fn = ~as.numeric(.))
serie_a_ranger_recipe %>%
prep(serie_a_training) %>%
juice()## # A tibble: 502 x 24
## position Int TklW OG PKcon Age MP Min Gls Ast Non_PK_G PK
## <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 CB 51 32 0 2 27 37 3258 2 0 2 0
## 2 RM 33 27 0 0 26 35 2516 12 8 10 2
## 3 LB 15 24 0 0 31 24 1807 3 2 3 0
## 4 CM 40 38 0 0 32 34 2741 0 0 0 0
## 5 CDM 28 23 0 0 21 25 1803 0 1 0 0
## 6 CB 18 20 0 1 22 32 2880 0 1 0 0
## 7 CB 25 26 0 1 31 24 1927 0 0 0 0
## 8 ST 9 15 0 2 28 33 2719 11 0 11 0
## 9 LB 15 24 0 0 31 24 1807 3 2 3 0
## 10 CM 37 52 0 0 32 28 1984 0 4 0 0
## # … with 492 more rows, and 12 more variables: PKatt <dbl>, CrdY <dbl>,
## # CrdR <dbl>, G_plus_A_per90 <dbl>, G_minus_Pk_per90 <dbl>,
## # G_plus_A_minus_PK_per90 <dbl>, Rk <dbl>, GF <dbl>, GA <dbl>, GD <dbl>,
## # Pts <dbl>, revision <fct>serie_a_ranger <- rand_forest(mtry = tune(),
min_n = tune(),
trees = 100) %>%
set_mode("classification") %>%
set_engine("ranger")
serie_a_ranger_wf <-
workflow() %>%
add_recipe(serie_a_ranger_recipe) %>%
add_model(serie_a_ranger)
serie_a_ranger_wf## ══ Workflow ════════════════════════════════════════════════════════════════════
## Preprocessor: Recipe
## Model: rand_forest()
##
## ── Preprocessor ────────────────────────────────────────────────────────────────
## 3 Recipe Steps
##
## ● step_rm()
## ● step_upsample()
## ● step_mutate_at()
##
## ── Model ───────────────────────────────────────────────────────────────────────
## Random Forest Model Specification (classification)
##
## Main Arguments:
## mtry = tune()
## trees = 100
## min_n = tune()
##
## Computational engine: rangerset.seed(494)
serie_a_cv <- vfold_cv(serie_a_training, v = 5)
serie_a_rf_grid <- grid_regular(min_n(), finalize(mtry(), serie_a_training %>% select(-revision)), levels = 3)
ctrl_res <- control_stack_grid()
serie_a_ranger_cv <- serie_a_ranger_wf %>%
tune_grid(resamples = serie_a_cv,
grid = serie_a_rf_grid,
control = ctrl_res)## ! Fold1: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold1: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold1: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold2: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold3: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold4: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 7/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 8/9: 31 columns were requested but there were 23...## ! Fold5: preprocessor 1/1, model 9/9: 31 columns were requested but there were 23...## # A tibble: 18 x 8
## mtry min_n .metric .estimator mean n std_err .config
## <int> <int> <chr> <chr> <dbl> <int> <dbl> <chr>
## 1 1 2 accuracy binary 0.908 5 0.0124 Preprocessor1_Model1
## 2 1 2 roc_auc binary 0.902 5 0.0138 Preprocessor1_Model1
## 3 1 21 accuracy binary 0.896 5 0.0140 Preprocessor1_Model2
## 4 1 21 roc_auc binary 0.888 5 0.0193 Preprocessor1_Model2
## 5 1 40 accuracy binary 0.901 5 0.0102 Preprocessor1_Model3
## 6 1 40 roc_auc binary 0.897 5 0.0181 Preprocessor1_Model3
## 7 16 2 accuracy binary 0.881 5 0.0194 Preprocessor1_Model4
## 8 16 2 roc_auc binary 0.879 5 0.0206 Preprocessor1_Model4
## 9 16 21 accuracy binary 0.869 5 0.0101 Preprocessor1_Model5
## 10 16 21 roc_auc binary 0.877 5 0.0200 Preprocessor1_Model5
## 11 16 40 accuracy binary 0.869 5 0.0114 Preprocessor1_Model6
## 12 16 40 roc_auc binary 0.880 5 0.0236 Preprocessor1_Model6
## 13 31 2 accuracy binary 0.867 5 0.0168 Preprocessor1_Model7
## 14 31 2 roc_auc binary 0.867 5 0.0246 Preprocessor1_Model7
## 15 31 21 accuracy binary 0.857 5 0.00866 Preprocessor1_Model8
## 16 31 21 roc_auc binary 0.868 5 0.0251 Preprocessor1_Model8
## 17 31 40 accuracy binary 0.859 5 0.0150 Preprocessor1_Model9
## 18 31 40 roc_auc binary 0.871 5 0.0218 Preprocessor1_Model9serie_a_best1 <- serie_a_ranger_cv %>%
select_best(metric = "accuracy")
serie_a_ranger_final_wf<- serie_a_ranger_wf %>%
finalize_workflow(serie_a_best1)set.seed(494)
serie_a_ranger_fit <- serie_a_ranger_final_wf %>%
fit(serie_a_training)
serie_a_rf_explain <-
explain_tidymodels(
model = serie_a_ranger_fit,
data = serie_a_training %>% select(-revision),
y = as.numeric(serie_a_training$revision == "TOTS"),
label = "rf"
)## Preparation of a new explainer is initiated
## -> model label : rf
## -> data : 412 rows 31 cols
## -> target variable : 412 values
## -> predict function : yhat.workflow will be used ( [33m default [39m )
## -> predicted values : No value for predict function target column. ( [33m default [39m )
## -> model_info : package tidymodels , ver. 0.1.3 , task classification ( [33m default [39m )
## -> predicted values : numerical, min = 0.001597412 , mean = 0.1840812 , max = 0.9732857
## -> residual function : difference between y and yhat ( [33m default [39m )
## -> residuals : numerical, min = -0.6253171 , mean = -0.0554404 , max = 0.7448798
## [32m A new explainer has been created! [39m
serie_a_ranger_test <- serie_a_ranger_final_wf %>%
last_fit(serie_a_split)
serie_a_ranger_test %>%
collect_metrics()## # A tibble: 2 x 4
## .metric .estimator .estimate .config
## <chr> <chr> <dbl> <chr>
## 1 accuracy binary 0.919 Preprocessor1_Model1
## 2 roc_auc binary 0.943 Preprocessor1_Model1## Truth
## Prediction Normal TOTS
## Normal 117 9
## TOTS 2 8serie_a_ranger_test <- serie_a_testing %>%
bind_cols(predict(serie_a_ranger_fit, new_data = serie_a_testing, type = "prob")) %>%
bind_cols(predict(serie_a_ranger_fit, new_data = serie_a_testing)) ## Truth
## Prediction Normal TOTS
## Normal 117 8
## TOTS 2 9## Player revision position Int TklW OG PKcon Nation Squad
## 1 Mattia Caldara 17 TOTS CB 90 36 0 0 ITA Atalanta
## 2 Giorgio Chiellini 18 TOTS CB 28 15 0 0 ITA Juventus
## 3 Federico Chiesa 18 TOTS RM 9 37 0 0 ITA Fiorentina
## 4 Edin Dzeko 18 Normal ST 2 10 0 0 BIH Roma
## 5 Fabio Quagliarella 18 TOTS ST 8 9 0 0 ITA Sampdoria
## 6 Emre Can 19 TOTS CM 21 58 1 1 GER Juventus
## 7 Giorgio Chiellini 19 TOTS CB 23 9 0 0 ITA Juventus
## 8 Rodrigo De Paul 19 TOTS CM 36 31 0 1 ARG Udinese
## 9 Mario Mandzukic 19 Normal ST 12 22 0 0 CRO Juventus
## 10 Allan 19 TOTS CM 16 92 0 0 BRA Napoli
## Age Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY
## 1 22 1994 30 2655 29.5 7 0 7 0 0 4
## 2 32 1984 26 2161 24.0 0 1 0 0 0 2
## 3 19 1997 36 3012 33.5 6 4 6 0 0 7
## 4 31 1986 36 3018 33.5 16 3 16 0 0 6
## 5 34 1983 35 2719 30.2 19 5 12 7 8 4
## 6 24 1994 29 1811 20.1 4 1 3 1 1 7
## 7 33 1984 25 1991 22.1 1 1 1 0 0 3
## 8 24 1994 36 3189 35.4 9 9 6 3 6 7
## 9 32 1986 25 2014 22.4 9 6 9 0 0 4
## 10 27 1991 33 2616 29.1 1 3 1 0 0 10
## CrdR G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0 0.24 0.00 0.24 0.24 0.24
## 2 0 0.00 0.04 0.04 0.00 0.04
## 3 0 0.18 0.12 0.30 0.18 0.30
## 4 0 0.48 0.09 0.57 0.48 0.57
## 5 0 0.63 0.17 0.79 0.40 0.56
## 6 0 0.20 0.05 0.25 0.15 0.20
## 7 0 0.05 0.05 0.09 0.05 0.09
## 8 0 0.25 0.25 0.51 0.17 0.42
## 9 0 0.40 0.27 0.67 0.40 0.67
## 10 0 0.03 0.10 0.14 0.03 0.14
## Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 4 62 41 21 72 16948 0.5831035 0.4168965 Normal
## 2 1 86 24 62 95 39316 0.8128594 0.1871406 Normal
## 3 8 54 46 8 57 26092 0.6652006 0.3347994 Normal
## 4 3 61 28 33 77 37450 0.4981857 0.5018143 TOTS
## 5 10 56 60 -4 54 20156 0.7249446 0.2750554 Normal
## 6 1 70 30 40 90 37799 0.6721292 0.3278708 Normal
## 7 1 70 30 40 90 37799 0.8219823 0.1780177 Normal
## 8 12 39 53 -14 43 20414 0.8304859 0.1695141 Normal
## 9 1 70 30 40 90 37799 0.4826763 0.5173237 TOTS
## 10 2 74 36 38 79 29003 0.7006194 0.2993806 Normalserie_a_modeling21 <- fifa21_modeling_serie_a %>%
mutate(revision = as.factor(revision), Nation = as.factor(Nation), Age = as.integer(Age))serie_a_modeling_outfield21 <- serie_a_modeling21 %>%
filter(position != "GK") %>%
#filter(position %in% c("ST", "LW", "RW", "CF", "CAM")) %>%
filter(minutes_played_divided_by90 >= 14) %>%
mutate(position = ifelse(position == "RWB", "RB", ifelse(position == "LWB", "LB", ifelse(position == "RW", "RM", ifelse(position == "LW", "LM", position)))))serie_a_ranger_test21 <- serie_a_modeling_outfield21 %>%
bind_cols(predict(serie_a_ranger_fit, new_data = serie_a_modeling_outfield21, type = "prob")) %>%
bind_cols(predict(serie_a_ranger_fit, new_data = serie_a_modeling_outfield21)) ## Warning: Novel levels found in column 'Nation': 'ARM', 'EQG', 'RUS', 'UKR',
## 'USA', 'WAL'. The levels have been removed, and values have been coerced to
## 'NA'.
## Warning: Novel levels found in column 'Nation': 'ARM', 'EQG', 'RUS', 'UKR',
## 'USA', 'WAL'. The levels have been removed, and values have been coerced to
## 'NA'.serie_a_ranger_test21 %>%
filter(position %in% c("ST", "RW", "CF", "LW")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Romelu Lukaku Normal ST 2 3 0 0 BEL Inter 27
## 2 Cristiano Ronaldo Normal ST 5 2 0 0 POR Juventus 36
## 3 Lautaro Martinez Normal ST 26 15 0 0 ARG Inter 23
## 4 Duvan Zapata Normal ST 9 6 0 0 COL Atalanta 30
## 5 Ciro Immobile Normal ST 4 9 0 0 ITA Lazio 31
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1993 32 2580 28.7 21 9 16 5 5 4 0
## 2 1985 29 2463 27.4 25 2 20 5 6 3 0
## 3 1997 33 2238 24.9 15 5 15 0 0 3 0
## 4 1991 32 2052 22.8 14 7 13 1 1 0 0
## 5 1990 30 2399 26.7 18 5 15 3 6 4 1
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.73 0.31 1.05 0.56 0.87
## 2 0.91 0.07 0.99 0.73 0.80
## 3 0.60 0.20 0.80 0.60 0.80
## 4 0.61 0.31 0.92 0.57 0.88
## 5 0.68 0.19 0.86 0.56 0.75
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 1 72 29 43 79 125 0.1496536 0.8503464 TOTS
## 2 Matches 3 65 30 35 66 118 0.2607361 0.7392639 TOTS
## 3 Matches 1 72 29 43 79 125 0.3807273 0.6192727 TOTS
## 4 Matches 2 78 39 39 68 118 0.3960695 0.6039305 TOTS
## 5 Matches 6 56 46 10 61 188 0.4556741 0.5443259 TOTSserie_a_ranger_test21 %>%
filter(position %in% c("CAM", "CM", "CDM", "LM", "RM")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age Born
## 1 Matteo Politano Normal RM 19 11 0 0 ITA Napoli 27 1993
## 2 Hirving Lozano Normal RM 13 18 0 0 MEX Napoli 25 1995
## 3 Lorenzo Insigne Normal LM 34 11 0 0 ITA Napoli 29 1991
## 4 Robin Gosens Normal LM 38 26 1 0 GER Atalanta 26 1994
## 5 Piotr Zielinski Normal CM 16 14 0 0 POL Napoli 26 1994
## MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 32 1696 18.8 9 4 9 0 0 3 0
## 2 27 1727 19.2 9 3 9 0 0 5 0
## 3 30 2415 26.8 17 6 10 7 7 2 1
## 4 27 2143 23.8 8 6 8 0 0 9 1
## 5 31 2154 23.9 6 8 6 0 0 2 0
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.48 0.21 0.69 0.48 0.69
## 2 0.47 0.16 0.63 0.47 0.63
## 3 0.63 0.22 0.86 0.37 0.60
## 4 0.34 0.25 0.59 0.34 0.59
## 5 0.25 0.33 0.58 0.25 0.58
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 4 73 37 36 66 125 0.4478159 0.5521841 TOTS
## 2 Matches 4 73 37 36 66 125 0.4575231 0.5424769 TOTS
## 3 Matches 4 73 37 36 66 125 0.4712953 0.5287047 TOTS
## 4 Matches 2 78 39 39 68 118 0.4928593 0.5071407 TOTS
## 5 Matches 4 73 37 36 66 125 0.4965284 0.5034716 TOTSserie_a_ranger_test21 %>%
filter(position %in% c("LB", "CB", "RB")) %>%
arrange(desc(.pred_TOTS)) %>%
head(5)## Player revision position Int TklW OG PKcon Nation Squad Age
## 1 Cristian Romero Normal CB 70 42 0 0 ARG Atalanta 23
## 2 Juan Cuadrado Normal RB 20 21 0 0 COL Juventus 32
## 3 Milan Skriniar Normal CB 26 26 0 0 SVK Inter 26
## 4 Rafael Toloi Normal CB 35 28 0 1 ITA Atalanta 30
## 5 Giovanni Di Lorenzo Normal RB 29 50 0 1 ITA Napoli 27
## Born MP Min minutes_played_divided_by90 Gls Ast Non_PK_G PK PKatt CrdY CrdR
## 1 1998 26 2095 23.3 2 2 2 0 0 10 0
## 2 1988 25 1812 20.1 0 10 0 0 0 8 1
## 3 1995 29 2507 27.9 3 0 3 0 0 1 0
## 4 1990 28 2283 25.4 2 0 2 0 0 7 0
## 5 1993 31 2790 31.0 2 5 2 0 0 11 0
## G_per90 A_per90 G_plus_A_per90 G_minus_Pk_per90 G_plus_A_minus_PK_per90
## 1 0.09 0.09 0.17 0.09 0.17
## 2 0.00 0.50 0.50 0.00 0.50
## 3 0.11 0.00 0.11 0.11 0.11
## 4 0.08 0.00 0.08 0.08 0.08
## 5 0.06 0.16 0.23 0.06 0.23
## Matches Rk GF GA GD Pts Attendance .pred_Normal .pred_TOTS .pred_class
## 1 Matches 2 78 39 39 68 118 0.5632139 0.4367861 Normal
## 2 Matches 3 65 30 35 66 118 0.6404613 0.3595387 Normal
## 3 Matches 1 72 29 43 79 125 0.6667806 0.3332194 Normal
## 4 Matches 2 78 39 39 68 118 0.6965193 0.3034807 Normal
## 5 Matches 4 73 37 36 66 125 0.7191976 0.2808024 Normal